The Agent Skills Directory

[COMMAND_EXECUTION]: Includes a Bash script in the Audit mode section of SKILL.md used to inventory skill files and extract metadata such as source and effort level. This is used for local reporting and does not involve network operations.
[PROMPT_INJECTION]: The skill processes potentially untrusted SKILL.md files during audit and review workflows, which presents a surface for indirect prompt injection. 1. Ingestion points: Reads SKILL.md and reference files from the local filesystem during review and audit tasks. 2. Boundary markers: No explicit delimiters or ignore instructions are used when reading data from other skills. 3. Capability inventory: The skill can write files, execute shell commands for inventorying, and delegate tasks to subagents via forward-testing. 4. Sanitization: No specific sanitization or validation of the content being reviewed is performed, relying on the agent's logic to maintain context.

skill-creator